1. Study of e+e- pp process using initial state radiation with BaBar
V. Druzhinin (BINP,Novosibirsk) for BaBar Collaboration

2. ISR method
Mass spectrum of pp system in the e+e-  pp reaction is related to cross section of e+e-  pp reaction at E=m.
e+e-  pp cross section depends on two form factors, electric GE and magnetic GM.
The ratio of form factors |GE/GM| can be obtained from the analysis of the proton angular distribution. The terms corresponding GM and GE have angular dependence close to 1+cos2 and sin2, respectively.
V.Druzhinin

3. Existing data
The existing data come from DCI, ADONE and BEPC e+e- experiments and from pp experiments at LEAR and Fermilab. All data were obtained under |GE|=|GM| assumption. The |GE/GM| ratio was measured at LEAR and found to be compatible to unity.
We are going to measure the e+e- pp from threshold up to 4.5 GeV. The advantage of ISR method over conventional e+e- and pp measurements is low dependence of the detection efficiency on mass and proton angle.
V.Druzhinin

4. BaBar detector DIRC Quartz Cherenkov radiator
Covers 80% of solid angle
Particle ID above 600 MeV/c
DCH
40 layers, axial and stereo wires.
Covers 92% of solid angle
dpt/pt ~ %
Particle ID up to 600 MeV/c
EMC:
6580 CsI(Tl) crystals
Covers 91% of solid angle
E resolution ~2 % at high E.
V.Druzhinin

5. Event selection At least 2 tracks with
Data: 232 fb-1 collected during
Process
Nexpect
pp 4k
+-
1000k
+-
2000k
K+K-
100k
At least 2 tracks with
Rxy < 2.5 cm
z < 6cm
p > 0.1 GeV/c
0.45 <  < 2.4
A photon with ECM > 3 GeV and <  < 2.4
Two tracks must pass proton selector
The kinematic fit is applied imposing energy and momentum conservation under different mass hypotheses
V.Druzhinin

6. Event selection K p p K
V.Druzhinin

7. K+K-, +-, +-, e+e- background
K+K- background contribution is estimated from the number of events with two identified protons but with 2K <20. The scale factor is obtained from the event sample with one identified photon and MC simulation.
The similar method is used for +- background.
The e+e- and +- background is estimated using the difference between their mass distributions and pp mass distribution for mass greater than 4.5 GeV.
data
+-
K+K-
+-
e+e-
4025
5.9 ± 2.5
2.5 ± 1.1
< 11
0.8 ± 0.8
V.Druzhinin

8. K+K-, +-, +-, e+e- background
data K  
V.Druzhinin

9. e+e-  pp 0 background
The most considerable source of background is the process e+e-pp0 with lost soft photon or with merged photons from 0 decay. These events have 2 distribution peaked at low 2 and can not be separated from the signal events.
The events of the e+e-pp0 process were selected using 2 of the kinematic fit to e+e-pp hypothesis and requirement on two-photon invariant mass. The mass spectrum of pp0 events passed our standard selection is determined as KMC(dN/dM)data. The scale factor found from simulation is about 3.
Mpp
<2.5
>4.5
Ndata
3166
322 57 4
Npp
171 ± 29
33 ± 11
25 ± 8
5 ± 3
V.Druzhinin

10. ISR background e+e-qq with other than pp0 final state
The main source of ISR background is e+e-pp0 process. Using the kinematic fit to pp0 hypothesis we select 847±31 events of this process. From the ratio of the detection efficiencies for pp and pp0 selections, 0.015±0.002, we estimate the number of e+e-pp0 events in pp event sample is 13±3.
The similar procedure is used for e+e-pp20 process. The 560±30 selected events of this process translate to 0.5 ±0.3 events passed pp selection.
N(1440)
e+e-qq with other than pp0 final state
is studied with JETSET simulation. Simulation yields 26±4 events. Two final states, pp20 and pp, dominate with 17 and 5 events, respectively.
V.Druzhinin

11. Background subtraction
+-
K+K-
pp0
pp0
uds
pp
data N1
5.9 ± 2.5
2.5 ± 1.0
229 ± 32
13 ± 3
26 ± 4
3737± 75
4025 N2
4.2 ± 1.8
1.3 ± 0.5
29 ± 5
20 ± 3
37 ± 5
179 ± 12
290 
0.71 ± 0.05
0.52 ± 0.04
0.13 ± 0.01
1.53 ± 0.25
1.44 ± 0.30
0.048 ± 0.003
N1 and N2 are numbers of events with 2 < 30 and 30 < 2 <60, respectively. =N1/N2
Since the result of JETSET for rare processes requires confirmation we use the method of background subtraction based on the difference of 2 distributions for signal and background events.
pp
Nbkg=50±20 is in an agreement with the estimation from simulation, 39±5.
pp0
pp0
V.Druzhinin

12. Mass spectrum
V.Druzhinin

13. Angular distribution
Mpp, GeV/c2 N
Nbkg
|GE/GM|
533
2±7
584
37±12
602
50±15
705
42±14
592
61±16
464
45±12
The distribution over the angle between the proton momentum in the pp rest frame and the momentum of pp system in the e+e- c.m. frame is fitted by the sum of histograms, obtained from two simulation event samples, one with GE=0 and the other with GM=0. These distributions are close to 1+cos2p and sin2p.
V.Druzhinin

14. Angular distribution
The simulated angular distribution are corrected for data-MC difference in PID , tracking and photon efficiencies.
About 10% variation of detection efficiency is explained by complex momentum dependence of PID efficiency which has minimum at momentum 1.5 GeV/c.
The sources of |GE/GM| systematic error:
background subtraction (0.16)
MC statistics (0.08)
|GE/GM| mass dependence (0.01)
efficiency correction (0.02)
V.Druzhinin

15. Angular distribution,cross-checks
To cross-check our method of |GE/GM| measurement we compare the data and simulated distributions over cos K for e+e-   K+K-  process.
We also study angular distribution for J/ pp decay which is described by 1+cos2p with =0.672±0.034.
Our result is
V.Druzhinin

16. |GE/GM| ratio
BaBar |GE/GM| measurements vs previous ones and dispersion relation prediction (yellow) based on JLab space-like GE/GM and analyticity
Hep:-ph/
DM2+FENICE
BaBar
LEAR
hep-ph/
E835
V.Druzhinin

17. Detection efficiency
To first approximation the detection efficiency is taken from simulation. The model dependence due to uncertainty in |GE/GM| ratio is about 1% for Mpp<3 GeV/c2, and 10% for higher masses where we use |GE|=|GM| assumption.
This MC efficiency must be corrected to account for data-MC difference in detector response:
V.Druzhinin

18. Efficiency corrections
Effect
1.9 GeV/c2
3.0 GeV/c2
4.5 GeV/c2
2 cuts
-0.7 ± 1.0
-1.1 ± 1.0
-1.7 ± 1.0
Track loss
0.8 ± 3.0
1.1 ± 3.0
1.0 ± 3.0
PID
2.5 ± 3.3
3.2 ± 2.4
3.5 ± 2.7
Photon inefficiency
-1.3 ± 0.1
Photon conversion
0.4 ± 0.2
Trigger
-0.6 ± 0.3 —
Total
1.1 ± 4.6
2.3 ± 4.0
1.9 ± 4.2
V.Druzhinin

19. Cross section
With chosen mass bin the resolution correction insignificantly changes the shape of mass spectrum but leads to about 20% increase in the errors and their correlation.
The radiative correction factor is calculated with use of generator level simulation. We evaluate the ratio of the mass spectrum generated with higher-order radiative correction included (structure function method) to pure Born mass spectrum. The theoretical uncertainty is about 1%.
V.Druzhinin

20. Cross section In reasonable agreement with e+e- previous measurements
Negative steps at M ~ 2.2 and 3 GeV (!?)
V.Druzhinin

21. Effective proton form factor
Steep behaviour at threshold confirmed
Similar behaviour at threshold is seen in other processes with different quantum numbers
V.Druzhinin

22. J/ and (2S) decays N ,% eeBpp, eV J/ 438 ± 22 17.2±0.7
12.0±0.6±0.5
(2S)
22.2 ± 5.5
16.5±0.7
0.70±0.17±0.03
(2S)
V.Druzhinin

23. Summary
The e+e- pp cross section and proton form factor have been measured from threshold up to 4.5 GeV.
The form factor have complex mass dependence. We confirm near-threshold enhancement observed in LEAR experiment. There are also two mass regions, near 2.25 and 3 GeV, with the rapid decrease of the form factor.
The |GE/GM| ratio has been measured from threshold to 3GeV. We have observe noticeable deviation of this ratio from unity, in disagreement with previous LEAR measurement.
V.Druzhinin